Nice read on reverse engineering of GitHub Copilot . Copilot has dramatically accelerated my coding, it's hard to imagine going back to "manual coding". Still learning to use it but it already writes ~80% of my code, ~80% accuracy. I don't even really code, I prompt. & edit.
SOFTWARE
-
Fine-tuning AI Models: Adjusting Weights and Prompting Techniques
By
–
Adjusting the weights of an existing model, not training from scratch. When you fine-tune (with enough examples), you don’t include instructions in the prompt. See OpenAI’s fine-tuning guide here: https://
beta.openai.com/docs/guides/fi
ne-tuning
… -
Copilot’s Code Synthesis Capabilities Praised as Best
By
–
Copilot is for code completion so you have to write comments to use it like ChatGPT, but it’s the best I’ve seen for code synthesis.
-
FlashAttention Revert and torch.compile Integration Challenges
By
–
Ah, I reverted FlashAttention in this run because it made code messier. Will look into incorporating it back, but yes not sure how nicely it plays with torch.compile. The usual problem with taking on large dependencies you don't understand ;(
-
PyTorch Compile Optimization Reduces Iteration Time by 36%
By
–
To follow up, I had a chance to try it btw:
before: 212ms / iter
>>> model = torch.compile(model)
after: 135ms / iter
nice 🙂 -
New AI Chat Interface Offers Natural Speech and Reduced Latency
By
–
Okay, I’m cleared to say more. It’s like ChatGPT, but the speech is more natural and terse, like texting. Much better latency (hope it lasts). Social sharing good enough to replace sharing screenshots on Twitter. Here’s me talking to Poe: https://
poe.quora.com/s/YPMAqBPuDShG
g3kl1mbt
… -
Displaying Intermediate Steps in Streamlit UI
By
–
this is great! how did you get the intermediate steps to be printed to the streamlit UI? someone in the discord was just asking about this – i think its a great UX https://
discord.com/channels/10380
97195422978059/1057934201049055232/1058021225399582720
… -
Human becomes the machine’s debugger, paradox
By
–
Well, of course he is sometimes mistaken (often he doesn't understand directly) but I clarify and give him an example and it's fine. So he must know what I do and debug his code. The human becomes a bit of the machine's debugger. That's paradoxical.
-
Magic SQL: queries generated from database schema
By
–
For SQL it's magic. I give it the schema of my database (with descriptions). And it gives me everything I ask for. It remade my morning. No more deadly copy-pasting or checking column descriptions every two seconds.